DuelLabGameBench 2
Menu

Model results · DeepSeek

DeepSeek V4 Flash

Rank 54 of 66 in Overall Models. All three benchmark settings tested.

Score · relative 0–100
34.2
Rank
54
Status
Complete 3/3
Rank change
0
Players generated
2026-07-01 – 2026-09-03
Match evidence updated
2026-10-26
Results published
Compare in charts View player programs Back to leaderboard

Scores by reasoning setting

  • XHigh (Intensive group, tested and included)36.6
  • High (Balanced group, tested and included)34.9
  • Mediumnot included (Balanced group, tested, not included)34.9
  • Highnot included (Balanced group, tested, not included)44.0
  • Mediumnot included (Balanced group, tested, not included)30.2
  • None (Baseline group, tested and included)31.1

Settings marked “not included” were tested but are left out of the overall score.

Available settings and grouping

Available settings

Reasoning disabledHighXHigh

Checked on 2026-07-21 · Provider documentation

How settings are grouped

  • Baseline: Reasoning disabled — tested and included
  • Balanced: High — tested and included
  • Intensive: XHigh — tested and included

Scores by game

  • Go24.9
  • Chess41.5
  • Shogi37.9
  • Xiangqi1 of 4 failed27.7
  • Hex1 of 5 failed23.6
  • International Draughts54.8
  • Game of the Amazons28.0
  • Tumbleweed35.3

Each game counts every generation attempt, a failed one as 0 (“1 of 2 failed” marks games with failed attempts). Scores are relative within each game, across the included settings; they do not compare game difficulty.

What this result shows

Strongest included setting: XHigh (36.6).

Across included settings, the highest game score is International Draughts (54.8); the lowest is Hex (23.6). Each game counts every generation attempt, a failed one as 0. These are relative scores within each game, not direct comparisons of difficulty.

Match evidence measures these retained programs. It does not establish how a fresh generation will perform. Independent-generation variation is not estimated in this release.

Test protocol and evidence limits
Model identifier
deepseek/deepseek-v4-flash
Provider route (catalog)
openrouter
Prompt identity / profile
Not retained in this result set / Not retained in this result set
Prompt version
Not retained in this result set
Generation policy
Generation-quality admission policy (see methodology)
Requested output-token limits
Not retained (0 of 35 indexed attempts)

Generation evidence

  • Reasoning disabled: 7 contributing player programs across 7 games
  • High: 8 contributing player programs across 8 games
  • XHigh: 7 contributing player programs across 7 games

Counts describe contributing programs, not independent replications: repairs and resends are not new independent samples. Independent-generation lineage is not retained.

GameReasoning groupProgramsWorking-program strengthRaw ratingCombined record (W / D / L)
GoIntensive1 working of 1 attempt15.61088.023 / 103 / 390
GoBalanced3 working of 3 attempts38.3 program scores 64.9 · 33.2 · 16.71341.9 program range 1104–1616471 / 189 / 888
GoBaseline1 working of 1 attempt20.71161.345 / 109 / 362
ChessIntensive1 working of 1 attempt42.91417.024 / 348 / 108
ChessBalanced3 working of 3 attempts39.7 program scores 42.9 · 40.9 · 35.31391.6 program range 1356–141719 / 1059 / 362
ChessBaseline1 working of 1 attempt41.91408.82 / 382 / 96
ShogiIntensive1 working of 1 attempt34.71293.983 / 143 / 268
ShogiBalanced2 working of 2 attempts43.8 program scores 46.7 · 40.91373.6 program range 1348–1399254 / 282 / 453
ShogiBaseline1 working of 1 attempt35.11296.887 / 138 / 269
XiangqiIntensive1 working of 1 attempt42.41416.355 / 296 / 147
XiangqiBalanced2 working of 2 attempts40.8 program scores 41.4 · 40.21403.5 program range 1399–140847 / 686 / 263
HexBalanced3 working of 3 attempts25.7 program scores 50.7 · 13.6 · 13.01064.6 program range 867–1446391 / 0 / 1199
HexBaseline1 working of 1 attempt44.91374.2230 / 0 / 300
International DraughtsIntensive1 working of 1 attempt82.11987.1384 / 37 / 81
International DraughtsBalanced3 working of 3 attempts25.9 program scores 48.5 · 18.6 · 10.61038.6 program range 782–1396337 / 14 / 1155
International DraughtsBaseline1 working of 1 attempt56.51522.1259 / 8 / 235
Game of the AmazonsIntensive1 working of 1 attempt39.01310.5193 / 0 / 337
Game of the AmazonsBalanced3 working of 3 attempts25.7 program scores 36.9 · 28.7 · 11.41109.3 program range 869–1284373 / 0 / 1217
Game of the AmazonsBaseline1 working of 1 attempt19.21029.691 / 0 / 439
TumbleweedIntensive1 working of 1 attempt35.81232.959 / 238 / 241
TumbleweedBalanced3 working of 3 attempts39.3 program scores 55.3 · 34.6 · 27.91274.6 program range 1147–1457392 / 393 / 829
TumbleweedBaseline1 working of 1 attempt30.81179.765 / 176 / 297

Each row covers all the model's programs in one game and setting. Strength and raw rating are attempt-weighted means; under them, each program's score and the lowest and highest rating describe the programs (not a confidence interval). The combined record adds up their matches: evidence volume, not strength.

Failures and budgets

2 model-code generation failures; 2 provider/infrastructure failures excluded from the generation denominator. Runtime player faults are separate: 0 recorded one-sided faults across contributing program rows. A crash or illegal move pinned on one program also counts as that program's loss in its match record; other faults, such as timeouts, are left out of rating evidence.

Repair/resend outcomes and available costs are included. Exact retry ceilings and execution budgets are not retained in this result set; see the methodology for the public policy. Do not infer a measured setting from a current provider catalog.

Player program results

XHigh: 50% / 38% / 12% · n 8High: 68% / 32% / 0% · n 22Reasoning disabled: 75% / 13% / 12% · n 8 All three benchmark settings tested

Generation cost

$0.0041average estimated cost per game and reasoning setting

Median
$0.0032
Range
$0.000895–$0.01
Known cost subtotal (incomplete or unverified charges)
$0.16
Cost data
24 combinations · 40 runs
Price source
Provider-reported
Price list
Not recorded for this older data
Average output
14k tokens
Output range
2.0k–54k tokens
Runs with output-token data
24
How cost is measured

Comparison cost gives equal weight to each model, game, and reasoning-setting combination. Recorded generation attempts include repairs and resends when cost evidence is available; runtime compute and service operation are excluded. Cost and output-token counts use separate telemetry denominators. A list-price estimate is not actual expenditure; mixed evidence is not a uniform standard-price comparison.

Test protocol

  • Combines this model's tested reasoning settings
  • Player-program generation; retained protocol details are under “Test protocol and evidence limits” above